Automated systems are available that can classify digital documents based on the statistical analysis of the content and data associated with the documents, while taking into consideration one or more user-defined rules. The rules are typically provided by a human operator to help enhance the classification capabilities of the automated system. Adding rules is a labor intensive task and although it may improve classification of some documents, it can also degrade the overall classification accuracy of the system, if a rule does not apply to some documents.